AI news and neural networks

RU / EN

/ AI Models

MistralAI releases Le Chonk: a model for cybersecurity and agents

Google Unveils Gemini 4 Carbon: A New Frontier in Secure, Agentic AI

NVIDIA Nemotron 3.5: Leading Open-Weight Model for Intelligent Multilingual Transcription

Comparison of GPT-6.1 Terra and Claude Opus 5.5 on a new benchmark

LlamaIndex launches OCRia-1 for advanced document analysis

Luma Labs AI boosts multimodal model efficiency with new attention mechanism

DeepSeek Releases V4.1-Flash: A Faster and More Powerful LLM

Perplexity integrates NVIDIA's Nemotron 3.5 Super VL model

Gemini 4 vs. Claude Opus: Testing AI in Soft-Body Physics Simulations

Google DeepMind Launches On-Device AI to Fight Scams While Protecting Privacy

Gemini 4 Argon Available in Cursor AI with Strong Benchmark Score

Google launches Gemini 4 Argon for solving complex real-world tasks

Open Multimodal Decision Models Set New Standards for Edge AI Efficiency

Perplexity AI releases multimodal embedding models for search

Google DeepMind released a multimodal embedding model for devices

Gemini introduces 'Skills' for automating repetitive tasks

OpenAI fixes image recognition error in GPT-6

OpenAI launches GPT-6 Sol and Luna: 50% cheaper and more powerful than competitors

Anthropic Unveils Claude Fable 5.1 and Restricted Mythos 5.1

Runway Unveils Solaris Model for Interface Generation

OpenAI Cuts GPT-5.6 Sol Prices by More Than 20%

OpenAI Accelerates GPT-5.6 Sol to 750 Tokens per Second

DeepSeek Releases V4-Pro for Complex Agentic Tasks

Google launches Gemini 3.7 Flash for coding and agents

OpenAI to Restrict Access to GPT-5.4 via ChatGPT

GPT-5.6 Targets Frontier AI at Lower Serving Cost

Updated 3 months ago

DeepSeek Launches Public Beta of V4-Flash API

NVIDIA Nemotron 3 Embed Leads RTEB Retrieval Rankings

Thinking Machines Releases Trillion-Parameter Inkling Model

OpenAI Brings GPT-5.6 to the API with Sol, Terra, and Luna